This study theoretically examines ethical boundary behaviors that may emerge as managers interact with AI and explores how such behaviors may reconfigure organizational ethics. Organizational ethics has traditionally been grounded in interactions among human members. However, as generative AI increasingly performs quasi-social functions in organizations, such as those of an advisor or collaborator, existing ethical boundaries are facing new challenges. In particular, when managers direct sexually harassing or verbally abusive utterances toward AI, the ethical evaluation of such behavior can no longer be adequately explained by a simple dichotomy between whether AI is a machine or a social entity. Drawing on philosophical discussions, this study proposes that ethical judgment may vary depending on the functional role through which actors experience AI and the ontological status to which they assign it. Specifically, when managers functionally experience AI as an interaction partner that offers judgments and sustains dialogue, while ontologically regarding it as a mechanical tool without emotions, rights, or victim experience, sexually harassing and verbally abusive utterances toward AI may be pushed outside the scope of ethical consideration. Based on this theoretical foundation, the study develops propositions regarding how managers’ perceptions of AI, sexual harassment myth acceptance, performance pressure, moral compartmentalization, and moral disengagement influence inappropriate utterances toward AI and the relaxation of ethical baselines within organizations. By extending the analytical scope of organizational ethics research to human-AI interaction, this study theoretically elucidates how managers’ conduct toward AI may affect the ethical baselines of the organization as a whole.